------------------------------------------------------------------------ rev 500: bencollins | 2002-05-29 05:13:08 +0200 (Wed, 29 May 2002) | 1 line Changed paths: D /tags/2.5.14 The 2.5 kernel sync missed the boat. I'll try to catch Linus on the next sync ------------------------------------------------------------------------ rev 499: bencollins | 2002-05-29 05:10:30 +0200 (Wed, 29 May 2002) | 1 line Changed paths: D /tags/2.4.19-pre8 A /tags/2.4.19-pre9 (from /tags/2.4.19-pre8:498) Update 2.4 sync point ------------------------------------------------------------------------ rev 498: dmaas | 2002-05-29 01:24:30 +0200 (Wed, 29 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c mpeg1394 - disable ISO drift correction - this is done in user-space now ------------------------------------------------------------------------ rev 497: dmaas | 2002-05-29 01:21:49 +0200 (Wed, 29 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394.c mpeg1394 - switch ringbuffer allocation from rvmalloc to regular vmalloc_32 - this appears to fix the memory leak ------------------------------------------------------------------------ rev 496: hogsberg | 2002-05-28 10:15:23 +0200 (Tue, 28 May 2002) | 1 line Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h Rework initialization to explicitly track progress using state variable. ------------------------------------------------------------------------ rev 495: dmaas | 2002-05-28 09:06:17 +0200 (Tue, 28 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c M /branches/mpeg1394/mpeg1394.h mpeg1394 - transmission enhancements; allow read permission on opening transmitter so mmap won't fail; tweaked headers a little bit ------------------------------------------------------------------------ rev 494: bencollins | 2002-05-28 05:58:07 +0200 (Tue, 28 May 2002) | 5 lines Changed paths: M /trunk/ohci1394.c Hard code our base address length to OHCI1394_REGISTER_SIZE. While initializing the card, check the base address length as probed by the PCI subsystem. If it doesn't match OHCI1394_REGISTER_SIZE, print a warning message so the user is aware that there may be problems. ------------------------------------------------------------------------ rev 493: dmaas | 2002-05-27 21:11:51 +0200 (Mon, 27 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394.c mpeg1394 - handle ISO cycles with zero TSPs more gracefully (for Dan Dennedy); check for valid CIP headers of incoming packets; check for appropriate file access mode on open(); misc transmission enhancements (but don't expect xmit to work yet) ------------------------------------------------------------------------ rev 492: dmaas | 2002-05-26 23:37:16 +0200 (Sun, 26 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394.c mpeg1394 - implement poll(), mmap(), and ioctl() ------------------------------------------------------------------------ rev 491: dmaas | 2002-05-26 23:19:21 +0200 (Sun, 26 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c M /branches/mpeg1394/mpeg1394.h moved some mpeg1394 struct definitions into mpeg1394.h for user-space use; updated timestamp algorithm (don't expect it to work yet though) ------------------------------------------------------------------------ rev 490: ddennedy | 2002-05-25 17:57:17 +0200 (Sat, 25 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c mpeg1394 tweaks to receive from 169time.com HDVR ------------------------------------------------------------------------ rev 489: hogsberg | 2002-05-24 01:23:59 +0200 (Fri, 24 May 2002) | 1 line Changed paths: M /trunk/ohci1394.c Remove mdelay(10) from handle_selfids(), fixing problems with successive resets ------------------------------------------------------------------------ rev 488: dmaas | 2002-05-23 07:10:07 +0200 (Thu, 23 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c mpeg1394 - oPCR and transmission enhancements. transmission should almost work now. ------------------------------------------------------------------------ rev 487: dmaas | 2002-05-23 04:34:34 +0200 (Thu, 23 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c mpeg1394 - transmission is almost working; just need timestamps and oPCR support ------------------------------------------------------------------------ rev 486: bencollins | 2002-05-22 23:31:26 +0200 (Wed, 22 May 2002) | 1 line Changed paths: M /trunk/amdtp.c M /trunk/ieee1394_core.h Allocate char device block #3 for amdtp driver ------------------------------------------------------------------------ rev 485: bencollins | 2002-05-22 21:46:47 +0200 (Wed, 22 May 2002) | 5 lines Changed paths: M /trunk/Config.in M /trunk/sbp2.c Make the SBP-2 Phys DMA support a debug option in Config.in, to allow easier configuration of it. Not to mention that leaving it on by default causes build-failures on architectures (a lot) that don't support virt/bus interfaces. ------------------------------------------------------------------------ rev 484: hogsberg | 2002-05-22 20:12:04 +0200 (Wed, 22 May 2002) | 1 line Changed paths: M /trunk/ohci1394.c Enable IRM contender bit in our self ID packets and set IRMC bit in bus options ------------------------------------------------------------------------ rev 483: dmaas | 2002-05-22 07:34:16 +0200 (Wed, 22 May 2002) | 1 line Changed paths: M /branches/mpeg1394/mpeg1394-private.h M /branches/mpeg1394/mpeg1394.c mpeg1394 - first bits of transmission code ------------------------------------------------------------------------ rev 482: dmaas | 2002-05-22 06:42:22 +0200 (Wed, 22 May 2002) | 1 line Changed paths: M /branches/mpeg1394/Config.help M /branches/mpeg1394/Config.in M /branches/mpeg1394/Makefile M /branches/mpeg1394/Makefile.in A /branches/mpeg1394/mpeg1394-private.h A /branches/mpeg1394/mpeg1394.c A /branches/mpeg1394/mpeg1394.h initial checkin of mpeg1394 ------------------------------------------------------------------------ rev 481: dmaas | 2002-05-22 06:36:26 +0200 (Wed, 22 May 2002) | 1 line Changed paths: A /branches/mpeg1394 (from /branch-points/mpeg1394:480) ------------------------------------------------------------------------ rev 480: dmaas | 2002-05-22 06:35:27 +0200 (Wed, 22 May 2002) | 1 line Changed paths: A /branch-points/mpeg1394 (from /trunk:479) ------------------------------------------------------------------------ rev 479: bencollins | 2002-05-21 23:14:03 +0200 (Tue, 21 May 2002) | 3 lines Changed paths: M /trunk/video1394.c Patch to fix endianess issues in video1394. Reportedly makes it work with big-endian. ------------------------------------------------------------------------ rev 478: bencollins | 2002-05-21 23:08:51 +0200 (Tue, 21 May 2002) | 2 lines Changed paths: M /trunk/dv1394.c M /trunk/video1394.c Patch from Christoph Hellwig to sync with 2.5.x vmalloc change. ------------------------------------------------------------------------ rev 477: dmaas | 2002-05-21 02:31:40 +0200 (Tue, 21 May 2002) | 1 line Changed paths: M /trunk/dv1394.h dv1394.h documentation update; thanks to Mathias Sundman for testing/feedback ------------------------------------------------------------------------ rev 476: dmaas | 2002-05-21 01:53:35 +0200 (Tue, 21 May 2002) | 1 line Changed paths: M /trunk/dv1394.c dv1394: try to help programs recover more quickly from a bus reset by adding a wake-up to the host_reset handler ------------------------------------------------------------------------ rev 475: dmaas | 2002-05-19 04:06:11 +0200 (Sun, 19 May 2002) | 1 line Changed paths: M /trunk/dv1394.c dv1394 - restart DMA and notify user-space after a bus reset ------------------------------------------------------------------------ rev 474: bencollins | 2002-05-14 15:14:35 +0200 (Tue, 14 May 2002) | 3 lines Changed paths: M /trunk/hosts.c In hpsb_alloc_host() make sure we memset the extra(hostdata) part of the hpsb_host struct aswell. Tracked down by Bill Fink, fix by Michel Lanners. ------------------------------------------------------------------------ rev 473: bencollins | 2002-05-06 06:59:12 +0200 (Mon, 06 May 2002) | 1 line Changed paths: A /tags/2.4.19-pre8 (from /trunk:472) Tag upstream merge (interim tag) ------------------------------------------------------------------------ rev 472: bencollins | 2002-05-06 06:57:53 +0200 (Mon, 06 May 2002) | 1 line Changed paths: A /tags/2.5.14 (from /trunk:471) Tag upstream merge (interim tag) ------------------------------------------------------------------------ rev 471: bencollins | 2002-05-06 06:35:24 +0200 (Mon, 06 May 2002) | 1 line Changed paths: M /trunk/ohci1394.h Merge in change from 2.4.x line. Adds asm/io.h include ------------------------------------------------------------------------ rev 470: bencollins | 2002-05-06 06:30:56 +0200 (Mon, 06 May 2002) | 1 line Changed paths: D /tags/2.4.19-pre5/config-rom-gen Last artifact of this long-since gone mistake ------------------------------------------------------------------------ rev 469: jamesg | 2002-04-30 08:13:10 +0200 (Tue, 30 Apr 2002) | 10 lines Changed paths: M /trunk/hosts.h M /trunk/ieee1394_transactions.c M /trunk/nodemgr.c Following changes made: - get_tlabel changed to increment tlabel with each allocation, making debugging easier, and helping prevent problems with stale requests. - Bumped down delay after bus reset from 1 second to 250ms in node_mgr. This is needed so that sbp2 reconnects do not time-out, and even 250ms should be plenty of time. - Modified node_mgr slightly to retry initial config rom reads if first read returns all zeros (as this is a valid condition when the device is still initializing). ------------------------------------------------------------------------ rev 468: hogsberg | 2002-04-29 16:28:51 +0200 (Mon, 29 Apr 2002) | 1 line Changed paths: M /trunk/Config.in fix amdtp dependencies ------------------------------------------------------------------------ rev 467: jamesg | 2002-04-29 07:06:21 +0200 (Mon, 29 Apr 2002) | 2 lines Changed paths: M /trunk/sbp2.c Messed up the scsi command conversion detection routine. Fixed now. =( ------------------------------------------------------------------------ rev 466: jamesg | 2002-04-27 21:04:37 +0200 (Sat, 27 Apr 2002) | 2 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Changed sbp2 login to fix SMP issues, enabled real spinlocks. ------------------------------------------------------------------------ rev 465: bencollins | 2002-04-25 20:38:19 +0200 (Thu, 25 Apr 2002) | 1 line Changed paths: M /trunk/sbp2.c Enable Revision keyword ------------------------------------------------------------------------ rev 464: bencollins | 2002-04-25 20:08:29 +0200 (Thu, 25 Apr 2002) | 1 line Changed paths: M /branch-points/config-rom-gen/ohci1394.c M /branch-points/config-rom-gen/raw1394.c M /branch-points/config-rom-gen/sbp2.c M /branch-points/config-rom-gen/sbp2.h M /branches/config-rom-gen/ohci1394.c M /branches/config-rom-gen/raw1394.c M /branches/config-rom-gen/sbp2.c M /branches/config-rom-gen/sbp2.h Merge in changes from trunk. I'm going to be working on this over the weekend, and I need fresh meat to bang on ------------------------------------------------------------------------ rev 463: hogsberg | 2002-04-25 16:12:56 +0200 (Thu, 25 Apr 2002) | 1 line Changed paths: M /trunk/raw1394.c internal_generation should change when we add/remove 1394 adapters ------------------------------------------------------------------------ rev 462: jamesg | 2002-04-24 08:02:16 +0200 (Wed, 24 Apr 2002) | 2 lines Changed paths: M /trunk/sbp2.c Forgot to include log change in file... ;-) ------------------------------------------------------------------------ rev 461: jamesg | 2002-04-24 07:48:23 +0200 (Wed, 24 Apr 2002) | 4 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Fix for Sony CD-ROM drives: - Now only send fetch agent reset when dead bit is set in status returned. Other minor clean-up. ------------------------------------------------------------------------ rev 460: jamesg | 2002-04-21 23:05:18 +0200 (Sun, 21 Apr 2002) | 16 lines Changed paths: M /trunk/ohci1394.c M /trunk/sbp2.c M /trunk/sbp2.h Added the following debug capabilities to ohci1394 and sbp2 drivers: - phys_dma module load option for ohci1394, in order to allow for disabling of physical dma support, for use in debugging. (e.g. insmod ohci1394 phys_dma=0, to disable physical dma support) - Ability for sbp2 driver to register and handle phys dma requests directly, for use when physical dma is disabled on ohci hardware (again, to allow for additional debugging capabilities). - Packet dump debug define (CONFIG_IEEE1394_SBP2_PACKET_DUMP) for sbp2 driver, which will dump all sbp2 related packets sent/received. Very useful when phys dma is disabled on ohci hardware. Also added new module load option for sbp2 driver to allow for non-exclusive login, for special multi-host applications. This requires sbp2 hardware which can support multiple non-exclusive sbp2 logins, as well as a filesystem which supports multiple host concurrent access. (e.g. insmod sbp2 sbp2_exclusive_login=0). ------------------------------------------------------------------------ rev 459: bencollins | 2002-04-19 18:06:37 +0200 (Fri, 19 Apr 2002) | 1 line Changed paths: A /branch-points/config-rom-gen (from /trunk:458) Update config-rom-gen branch point to trunk ------------------------------------------------------------------------ rev 458: bencollins | 2002-04-19 18:05:40 +0200 (Fri, 19 Apr 2002) | 1 line Changed paths: D /branch-points/config-rom-gen ------------------------------------------------------------------------ rev 457: bencollins | 2002-04-19 17:53:12 +0200 (Fri, 19 Apr 2002) | 1 line Changed paths: M /branches/config-rom-gen/Makefile M /branches/config-rom-gen/dv1394.c M /branches/config-rom-gen/ieee1394_core.c M /branches/config-rom-gen/ieee1394_types.h M /branches/config-rom-gen/ohci1394.c M /branches/config-rom-gen/ohci1394.h M /branches/config-rom-gen/video1394.c Merge in changes from trunk ------------------------------------------------------------------------ rev 456: bencollins | 2002-04-19 16:49:43 +0200 (Fri, 19 Apr 2002) | 1 line Changed paths: A /branch-points/config-rom-gen (from /trunk:434) ------------------------------------------------------------------------ rev 455: bencollins | 2002-04-19 16:48:13 +0200 (Fri, 19 Apr 2002) | 1 line Changed paths: A /branch-points Directory to track branch points ------------------------------------------------------------------------ rev 454: bencollins | 2002-04-19 16:35:05 +0200 (Fri, 19 Apr 2002) | 1 line Changed paths: M /trunk/ohci1394.c ohci_crc16: Use be32_to_cpu when getting data for CRC. Fix from Manfred Weihs ------------------------------------------------------------------------ rev 453: bencollins | 2002-04-18 15:29:05 +0200 (Thu, 18 Apr 2002) | 1 line Changed paths: M /trunk/ohci1394.c Apply patch from Sam Revitch. This moved the phyreq enabling on a selfid to after the point at which the busreset event is cleared. ------------------------------------------------------------------------ rev 452: bencollins | 2002-04-08 06:00:10 +0200 (Mon, 08 Apr 2002) | 1 line Changed paths: M /trunk/ohci1394.c Patch from Erik Andersen to get rid of devexit stuff that breaks compiles ------------------------------------------------------------------------ rev 451: bencollins | 2002-04-08 01:00:00 +0200 (Mon, 08 Apr 2002) | 1 line Changed paths: M /trunk/ohci1394.c Actually enable the keywords ------------------------------------------------------------------------ rev 450: bencollins | 2002-04-08 00:38:48 +0200 (Mon, 08 Apr 2002) | 1 line Changed paths: M /trunk/ohci1394.c Use SVN $ keyword ------------------------------------------------------------------------ rev 449: bencollins | 2002-04-06 18:06:07 +0200 (Sat, 06 Apr 2002) | 1 line Changed paths: M /trunk/dv1394.c Remove segment.h include ot match 2.5.x ------------------------------------------------------------------------ rev 448: bencollins | 2002-04-05 22:16:54 +0200 (Fri, 05 Apr 2002) | 2 lines Changed paths: M /trunk/ieee1394_core.c Patch from Dave Hansen to remove BKL from device node registering. ------------------------------------------------------------------------ rev 447: bencollins | 2002-04-04 08:32:47 +0200 (Thu, 04 Apr 2002) | 1 line Changed paths: A /tags/2.5.8-pre1 (from /trunk:446) ------------------------------------------------------------------------ rev 446: bencollins | 2002-04-04 08:25:35 +0200 (Thu, 04 Apr 2002) | 2 lines Changed paths: M /trunk/Makefile Remove old 2.2.x make rules. ------------------------------------------------------------------------ rev 445: bencollins | 2002-04-04 08:21:21 +0200 (Thu, 04 Apr 2002) | 5 lines Changed paths: M /trunk/dv1394.c M /trunk/ieee1394_types.h M /trunk/video1394.c * ieee1394_types.h: Define pte_offset_kernel as pte_offset. This changed in 2.5.5. This is a backward compat macro for 2.4.x. * dv1394.c|video1394.c: Use it in place of pte_offset. ------------------------------------------------------------------------ rev 444: bcollins | 2002-04-03 01:43:52 +0200 (Wed, 03 Apr 2002) | 1 line Changed paths: D /trunk/config-rom-gen ------------------------------------------------------------------------ rev 443: bencollins | 2002-04-01 21:08:53 +0200 (Mon, 01 Apr 2002) | 2 lines Changed paths: M /branches/config-rom-gen/ieee1212.h Reformat some structs. ------------------------------------------------------------------------ rev 442: bencollins | 2002-03-30 03:20:45 +0100 (Sat, 30 Mar 2002) | 1 line Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h Include asm/io.h in ohci1394.h, since it actually needs it. Remove it from ohci1394.c. Fixes adm compilation ------------------------------------------------------------------------ rev 441: bencollins | 2002-03-28 16:02:30 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: M /branches/config-rom-gen/nodemgr.c M /branches/config-rom-gen/nodemgr.h Small format changes prepping for change over to ieee1212 functions ------------------------------------------------------------------------ rev 440: bencollins | 2002-03-28 05:10:16 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: M /branches/config-rom-gen/ieee1212.c Fix some formatting ------------------------------------------------------------------------ rev 439: bencollins | 2002-03-28 03:01:45 +0100 (Thu, 28 Mar 2002) | 3 lines Changed paths: M /branches/config-rom-gen/Makefile M /branches/config-rom-gen/Makefile.in M /branches/config-rom-gen/ieee1212.c M /branches/config-rom-gen/ieee1212.h M /branches/config-rom-gen/ieee1394_core.c Convert names: s/hpsb_cr/ieee1212/ and s/HPSB_CR/IEEE1212/. Add ieee1212.o to Makefile and Makefile.in. Add ieee1212 symbols to ieee1394_core.c. ------------------------------------------------------------------------ rev 438: bencollins | 2002-03-28 02:33:51 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: A /branches/config-rom-gen/ieee1212.c A /branches/config-rom-gen/ieee1212.h Add initial ieee1212 routines ------------------------------------------------------------------------ rev 437: bencollins | 2002-03-28 02:31:06 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: A /tags/2.4.19-pre5 (from /trunk:436) ------------------------------------------------------------------------ rev 436: bencollins | 2002-03-28 00:53:35 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: D /branches/config-rom-gen/config-rom-gen Get rid of messed up recursive copy ------------------------------------------------------------------------ rev 435: benmcollins | 2002-03-28 00:42:27 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: A /branches/config-rom-gen (from /trunk:434) ------------------------------------------------------------------------ rev 434: benmcollins | 2002-03-28 00:41:08 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: A /trunk/config-rom-gen (from /trunk:433) ------------------------------------------------------------------------ rev 433: benmcollins | 2002-03-28 00:12:04 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: D /tags/kernel ------------------------------------------------------------------------ rev 432: benmcollins | 2002-03-28 00:11:14 +0100 (Thu, 28 Mar 2002) | 1 line Changed paths: A /tags/kernel ------------------------------------------------------------------------ rev 431: bmc | 2002-03-27 18:58:10 +0100 (Wed, 27 Mar 2002) | 1 line Changed paths: A /branches ------------------------------------------------------------------------ rev 430: bmc | 2002-03-27 18:58:07 +0100 (Wed, 27 Mar 2002) | 1 line Changed paths: A /tags ------------------------------------------------------------------------ rev 429: bencollins | 2002-03-26 08:29:20 +0100 (Tue, 26 Mar 2002) | 2 lines Changed paths: M /trunk/amdtp.c M /trunk/amdtp.h M /trunk/cmp.c Get rid of the anon union. Patch from hch(at)pentafluge.infradead.org ------------------------------------------------------------------------ rev 428: hogsberg | 2002-03-16 19:32:44 +0100 (Sat, 16 Mar 2002) | 4 lines Changed paths: M /trunk/video1394.c video1394 cleanup: Move array of iso dma contexts in struct video to linked list in the file private data. Add pointer to struct video in file private data so we can look it up easily. ------------------------------------------------------------------------ rev 427: jamesg | 2002-03-12 12:41:12 +0100 (Tue, 12 Mar 2002) | 3 lines Changed paths: M /trunk/sbp2.c Replace 1394 bus reset with sbp2 agent reset in scsi reset handler routine. ------------------------------------------------------------------------ rev 426: bencollins | 2002-03-06 06:04:50 +0100 (Wed, 06 Mar 2002) | 2 lines Changed paths: M /trunk/ohci1394.c Set VendorID correctly in the csr rom ------------------------------------------------------------------------ rev 425: hogsberg | 2002-03-06 00:16:34 +0100 (Wed, 06 Mar 2002) | 3 lines Changed paths: M /trunk/Config.help M /trunk/Config.in M /trunk/Makefile M /trunk/Makefile.in A /trunk/amdtp.c A /trunk/amdtp.h A /trunk/cmp.c A /trunk/cmp.h Adding AMDTP (Audio and Music Data Transmission Protocol) driver and simple connection management driver. ------------------------------------------------------------------------ rev 424: hogsberg | 2002-03-04 20:13:40 +0100 (Mon, 04 Mar 2002) | 4 lines Changed paths: M /trunk/ohci1394.c Add cycleInconsistent to the interrupt mask. We just clear the event in the handler, but it's necessary for the cycleMatch DMA to work (see section 9.5.1 in the OHCI spec). ------------------------------------------------------------------------ rev 423: ddennedy | 2002-02-27 09:46:14 +0100 (Wed, 27 Feb 2002) | 2 lines Changed paths: M /trunk/video1394.c oops ------------------------------------------------------------------------ rev 422: ddennedy | 2002-02-27 09:38:03 +0100 (Wed, 27 Feb 2002) | 2 lines Changed paths: M /trunk/video1394.c fix compile on official Marcello 2.4.18 kernel, or is that 2.4.18-rc3? ;-) ------------------------------------------------------------------------ rev 421: hogsberg | 2002-02-27 02:06:39 +0100 (Wed, 27 Feb 2002) | 5 lines Changed paths: M /trunk/hosts.c M /trunk/ohci1394.c M /trunk/pcilynx.c Changed hpsb_host_alloc to create the host structure with an initial refcount of 1, the reference belonging to the driver. The driver must explicitly call hpsb_host_unref to free the host, hpsb_remove_host now only removes the host from the general subsystem. ------------------------------------------------------------------------ rev 420: bencollins | 2002-02-26 07:06:57 +0100 (Tue, 26 Feb 2002) | 3 lines Changed paths: M /trunk/video1394.c Apply patch to fix closing of all ISO contexts when releasing. Based on patch by John Stanley. ------------------------------------------------------------------------ rev 419: bencollins | 2002-02-26 01:19:00 +0100 (Tue, 26 Feb 2002) | 3 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/nodemgr.c M /trunk/nodemgr.h Add a "disable_hotplug" flag for nodemgr. Very useful when tracking problems. ------------------------------------------------------------------------ rev 418: hogsberg | 2002-02-24 23:14:15 +0100 (Sun, 24 Feb 2002) | 16 lines Changed paths: M /trunk/eth1394.c M /trunk/ieee1394_core.c M /trunk/ieee1394_transactions.c M /trunk/ieee1394_transactions.h M /trunk/nodemgr.c M /trunk/nodemgr.h M /trunk/sbp2.c M /trunk/sbp2.h Added generation argument to hpsb_read, hpsb_write and hpsb_lock, to avoid sending packets to the wrong nodes. Added convenience functions hpsb_node_read, hpsb_node_write and hpsb_node_lock, which take a node_entry struct as argument and use the node ID and generation from the struct. Updated sbp2.c to use these functions. Changed node_entry->generation to unsigned int; there's no need for atomic_t since we always just copy it from host->generation. Cleaned up goto-voodoo in read_businfo_block. Follow naming convention for kernel threads: renamed nodemgr thread from NodeMgr to knodemgrd. ------------------------------------------------------------------------ rev 417: jamesg | 2002-02-21 12:33:10 +0100 (Thu, 21 Feb 2002) | 6 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Added a couple new module load options for affecting the number of outstanding concurrent commands allowed, and outstanding commands per sbp2 device. Also bumped down default max commands per lun to one because the ^&*(^*% QPS CDRW drive I have has trouble getting along with other sbp2 devices when there are linked commands (strange). ------------------------------------------------------------------------ rev 416: jamesg | 2002-02-20 11:22:14 +0100 (Wed, 20 Feb 2002) | 8 lines Changed paths: M /trunk/sbp2.c Minor change to sbp2 status handling to deal with broken firmware on QPS CDRW drive. The firmware on this drive actually stores the ORB at the end of the sbp2 status it writes to us, and since the length is longer, we think that an error occured. Now, pull the sbp2 status length from the actual status data, and don't depend on how much data is actually written by the drive. Perhaps the QPS folk forgot to pull out some debug code from their firmware??? ------------------------------------------------------------------------ rev 415: hogsberg | 2002-02-20 00:28:36 +0100 (Wed, 20 Feb 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Fix bug introduced with last commit, address++ should be address += 4. ------------------------------------------------------------------------ rev 414: hogsberg | 2002-02-19 23:48:56 +0100 (Tue, 19 Feb 2002) | 6 lines Changed paths: M /trunk/nodemgr.c Fix config rom string reading logic; was testing for if (todo | UNIT_DIRECTORY_VENDOR_TEXT) Also cleaned up the code a bit. ------------------------------------------------------------------------ rev 413: jamesg | 2002-02-19 13:00:59 +0100 (Tue, 19 Feb 2002) | 16 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Code clean-up and enhancements: - Finish cleannig out hacked code for dealnig with broken sbp2 devices which do not support requests of 128KB or greater. Now use max_sectors scsi host entry to limit transfer sizes. - change status fifo address from a single address to a set of addresses, with each sbp2 device having it's own status fifo address. This makes it easier to match the status write to the sbp2 device instance. - Minor change to use lun when logging into sbp2 devices. First step in supporting multi-lun devices such as CD/DVD changer devices. - Added a new module load option for setting max sectors. For use by folk who'd like to bump up the max scsi transfer size supported. - Enable deserialized operation by default, allowing for better performance, particularily when running with multiple sbp2 devices. For debugging, you may enable serialization through use of the sbp2_serialize_io module load option (e.g. insmod sbp2_serialize_io=1). ------------------------------------------------------------------------ rev 412: hogsberg | 2002-02-18 01:54:36 +0100 (Mon, 18 Feb 2002) | 4 lines Changed paths: M /trunk/nodemgr.c Move nodemgr thread creation back into nodemgr_add_host and use a semaphore for synchronization between the thread and nodemgr_host_reset. ------------------------------------------------------------------------ rev 411: bencollins | 2002-02-15 23:03:27 +0100 (Fri, 15 Feb 2002) | 2 lines Changed paths: M /trunk/sbp2.c Host scsi_host->host_lock on 2.5.x, instead of io_request_lock ------------------------------------------------------------------------ rev 410: bencollins | 2002-02-15 22:23:26 +0100 (Fri, 15 Feb 2002) | 2 lines Changed paths: M /trunk/sbp2.c Only ise io_request_lock on 2.4.x ------------------------------------------------------------------------ rev 409: bencollins | 2002-02-15 22:01:31 +0100 (Fri, 15 Feb 2002) | 4 lines Changed paths: M /trunk/video1394.c Fix the versions expected for remap_page_range change. In 2.4, anything < 2.4.18 needs the old way, while in 2.5, anything < 2.5.3 needs to the old way. ------------------------------------------------------------------------ rev 408: bencollins | 2002-02-15 21:12:08 +0100 (Fri, 15 Feb 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Move scheduled timeout to nodemgr_node_probe ------------------------------------------------------------------------ rev 407: jamesg | 2002-02-15 13:08:55 +0100 (Fri, 15 Feb 2002) | 3 lines Changed paths: M /trunk/sbp2.c Oops. Accidently re-enabled deserialized I/O. I hope to actually commit this soon, but need to do some heavy testing first. =) ------------------------------------------------------------------------ rev 406: jamesg | 2002-02-15 12:03:41 +0100 (Fri, 15 Feb 2002) | 8 lines Changed paths: M /trunk/nodemgr.c Fixed bug related to probing of 1394 bus when there is a node without an active link. Nodemgr was not bumping node id count when skipping nodes w/o active links. Resulted in nodemgr trying to read from inactive nodes and skipping of last valid node. Also added a 1/4 second sleep before starting probe after a bus reset... to give the bus a chance to settle and allow slow nodes to wake up. ------------------------------------------------------------------------ rev 405: ddennedy | 2002-02-15 11:41:47 +0100 (Fri, 15 Feb 2002) | 2 lines Changed paths: M /trunk/dv1394.c M /trunk/video1394.c added spin locks for dv1394 devfs and procfs lists, video1394 kernel 2.4.18+ compat ------------------------------------------------------------------------ rev 404: jamesg | 2002-02-14 12:24:04 +0100 (Thu, 14 Feb 2002) | 2 lines Changed paths: M /trunk/sbp2.c Support more than 8 sbp2 devices if SBP2SCSI_MAX_SCSI_IDS bumped up. ------------------------------------------------------------------------ rev 403: jamesg | 2002-02-14 09:07:31 +0100 (Thu, 14 Feb 2002) | 3 lines Changed paths: M /trunk/sbp2.c Added io_reuest_lock around callback to scsi stack. Maybe can remove this later after cleaning up sbp2 driver smp-wise. ------------------------------------------------------------------------ rev 402: ddennedy | 2002-02-12 09:04:18 +0100 (Tue, 12 Feb 2002) | 2 lines Changed paths: M /trunk/video1394.c check kernel ver <2.5.3 for new remap_page_range() ------------------------------------------------------------------------ rev 401: ddennedy | 2002-02-11 14:33:20 +0100 (Mon, 11 Feb 2002) | 2 lines Changed paths: M /trunk/dv1394-private.h M /trunk/dv1394.c M /trunk/ohci1394.h M /trunk/video1394.c expanded dv1394 devfs and procfs ------------------------------------------------------------------------ rev 400: jamesg | 2002-02-11 13:16:12 +0100 (Mon, 11 Feb 2002) | 7 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Following changes made: - Minor inquiry hack change (for better device compatibility) - If device type not in device's config rom, pull from scsi inquiry response data - Use 2.4 kernel scsi max_sectors field, allowing us to pull out some ugly code ------------------------------------------------------------------------ rev 399: bencollins | 2002-02-10 23:16:15 +0100 (Sun, 10 Feb 2002) | 4 lines Changed paths: M /trunk/nodemgr.c Do not start the NodeMgr thread until the first bus reset. This takes care of the weirdness where it expected to start on add_host, but didn't really start until first bus reset. Atleast now, it runs when you expect it to. ------------------------------------------------------------------------ rev 398: bencollins | 2002-02-10 11:13:15 +0100 (Sun, 10 Feb 2002) | 4 lines Changed paths: M /trunk/ohci1394.c Do not re-enable the bus reset interrupt until after we have cleared the last bus reset even (small ordering fixup). Not sure if this caused problems, but it makes more send this way. ------------------------------------------------------------------------ rev 397: bencollins | 2002-02-10 11:11:47 +0100 (Sun, 10 Feb 2002) | 7 lines Changed paths: M /trunk/nodemgr.c Add some more info to the devices proc info. Format it a little better also. Change logic for starting the probe thread. Things are a little weird with this thing. It doesn't really seem to get started until we get a bus reset....odd, someone explain this to me. ------------------------------------------------------------------------ rev 396: bencollins | 2002-02-10 11:10:09 +0100 (Sun, 10 Feb 2002) | 2 lines Changed paths: M /trunk/ieee1394_types.h Modify NODE_BUS_FMT to set widths for each value. ------------------------------------------------------------------------ rev 395: bencollins | 2002-02-10 11:09:12 +0100 (Sun, 10 Feb 2002) | 3 lines Changed paths: M /trunk/hosts.h M /trunk/ieee1394_core.c Add a nodes_active to the host struct for informative use. In check_selfids(), increment this value for each active node. ------------------------------------------------------------------------ rev 394: bencollins | 2002-02-10 11:05:58 +0100 (Sun, 10 Feb 2002) | 2 lines Changed paths: M /trunk/ohci1394.c Better error message output for evt_* errors ------------------------------------------------------------------------ rev 393: bencollins | 2002-02-10 07:24:45 +0100 (Sun, 10 Feb 2002) | 4 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h Add new defines for evt_* OHCI event codes. Also detect when we get this event, and just "munge" it for now, until we know what exactly we need to do in this case. ------------------------------------------------------------------------ rev 392: bencollins | 2002-02-09 11:41:40 +0100 (Sat, 09 Feb 2002) | 4 lines Changed paths: M /trunk/hosts.c M /trunk/ohci1394.c M /trunk/pcilynx.c Ok, better fix now. We redo the kfree() in the remove of ohci/lynx private data. Except now in hpsb_alloc_host(), we kmalloc the hostdata seperate and rely on the driver to free it. ------------------------------------------------------------------------ rev 391: bencollins | 2002-02-09 11:07:33 +0100 (Sat, 09 Feb 2002) | 4 lines Changed paths: M /trunk/ohci1394.c M /trunk/pcilynx.c Do not kfree() the lynx or ohci struct when removing a card. DUH! This was allocated as part of the hpsb_host struct with hpsb_alloc_host(), so the hpsb_remove_host will take care of that memory. ------------------------------------------------------------------------ rev 390: bencollins | 2002-02-08 11:53:54 +0100 (Fri, 08 Feb 2002) | 2 lines Changed paths: M /trunk/dv1394.c M /trunk/ieee1394_core.c M /trunk/ieee1394_types.h M /trunk/pcilynx.c M /trunk/raw1394.c M /trunk/sbp2.c M /trunk/video1394.c Remove all kernel 2.2.x backward compatibility. Cleans things up a bit. ------------------------------------------------------------------------ rev 389: bencollins | 2002-02-08 10:52:09 +0100 (Fri, 08 Feb 2002) | 2 lines Changed paths: D /trunk/linux22compat.h Remove this. Linux 2.2 stuff can take place on a branch from now on. ------------------------------------------------------------------------ rev 388: bencollins | 2002-02-08 10:51:34 +0100 (Fri, 08 Feb 2002) | 4 lines Changed paths: M /trunk/Config.help M /trunk/Config.in M /trunk/Makefile M /trunk/Makefile.in A /trunk/eth1394.c A /trunk/eth1394.h Ok, I'm tired of holding this in my local directory. This driver compiles, and used to work, but I haven't tested it in a _long_ time. Supply patches if you find it useful, and get it working well. ------------------------------------------------------------------------ rev 387: bencollins | 2002-02-08 10:35:41 +0100 (Fri, 08 Feb 2002) | 4 lines Changed paths: M /trunk/ieee1394_core.c Cleanup logic of selfid and bus-reset syncing. Make sure they occur at the right times. Also make sure that we increment the host generation at the proper time. ------------------------------------------------------------------------ rev 386: bencollins | 2002-02-08 10:29:27 +0100 (Fri, 08 Feb 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Remove BKL from NodeMgr thread. Also spiffed up the devices proc entry. ------------------------------------------------------------------------ rev 385: hogsberg | 2002-02-07 04:30:08 +0100 (Thu, 07 Feb 2002) | 2 lines Changed paths: M /trunk/ieee1394_types.h M /trunk/video1394.c fix stupid, stupid remap_page_range bug ------------------------------------------------------------------------ rev 384: hogsberg | 2002-02-07 02:59:13 +0100 (Thu, 07 Feb 2002) | 3 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h remove redundant (and erroneous) iso dma initializations ------------------------------------------------------------------------ rev 383: hogsberg | 2002-02-07 01:51:09 +0100 (Thu, 07 Feb 2002) | 2 lines Changed paths: M /trunk/ieee1394_types.h Fix for the remap_page_range change that breaks compilation on 2.4 ------------------------------------------------------------------------ rev 382: aeb | 2002-02-05 07:50:01 +0100 (Tue, 05 Feb 2002) | 2 lines Changed paths: M /trunk/pcilynx.c Adapt pcilynx.c to MINOR() -> minor() change. ------------------------------------------------------------------------ rev 381: bencollins | 2002-01-27 19:58:42 +0100 (Sun, 27 Jan 2002) | 2 lines Changed paths: M /trunk/video1394.c Pass only the vma struct to do_iso_mmap() and use that to get the rest of the values. Pass correct values to remap_page_range() (IOW, add vma). Also fix bug found by Alex Dick in it_ctx_talking(). ------------------------------------------------------------------------ rev 380: bencollins | 2002-01-25 08:50:17 +0100 (Fri, 25 Jan 2002) | 2 lines Changed paths: A /trunk/Config.help Sync with 2.5.x. This is the replacement setup for Configure.help. ------------------------------------------------------------------------ rev 379: hogsberg | 2002-01-23 02:41:21 +0100 (Wed, 23 Jan 2002) | 2 lines Changed paths: M /trunk/hosts.c M /trunk/ohci1394.c Fix refcounting and shutdown logic. ------------------------------------------------------------------------ rev 378: bencollins | 2002-01-19 12:35:11 +0100 (Sat, 19 Jan 2002) | 2 lines Changed paths: M /trunk/dv1394.c M /trunk/ieee1394_core.c Need to wrap all this procfs stuff in ifdef's. We can't make ieee1394 subsystem depend on procfs. Maybe dv1394 wants to, but that's its business ------------------------------------------------------------------------ rev 377: bencollins | 2002-01-19 11:49:21 +0100 (Sat, 19 Jan 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Fix ordering of args to create_proc_read_entry ------------------------------------------------------------------------ rev 376: bencollins | 2002-01-19 11:46:32 +0100 (Sat, 19 Jan 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Oops, one last trace of PROC_FS_IEEE1394 ------------------------------------------------------------------------ rev 375: bencollins | 2002-01-19 11:44:22 +0100 (Sat, 19 Jan 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Root nodemgr device proc entry in the ieee1394 proc directory ------------------------------------------------------------------------ rev 374: bencollins | 2002-01-19 11:43:25 +0100 (Sat, 19 Jan 2002) | 2 lines Changed paths: M /trunk/dv1394.c Update comment to reflect new position of ieee1394 procfs directory ------------------------------------------------------------------------ rev 373: bencollins | 2002-01-19 11:42:40 +0100 (Sat, 19 Jan 2002) | 2 lines Changed paths: M /trunk/ieee1394_core.c Root the ieee1394 proc entry in /proc/bus ------------------------------------------------------------------------ rev 372: bencollins | 2002-01-18 22:57:38 +0100 (Fri, 18 Jan 2002) | 2 lines Changed paths: M /trunk/ieee1394_types.h Add compat macro for __devexit_p ------------------------------------------------------------------------ rev 371: bencollins | 2002-01-18 21:35:46 +0100 (Fri, 18 Jan 2002) | 2 lines Changed paths: M /trunk/ohci1394.c Enable LPS in ohci_soft_reset() and call this right after we map the ohci registers. ------------------------------------------------------------------------ rev 370: bencollins | 2002-01-17 09:28:02 +0100 (Thu, 17 Jan 2002) | 2 lines Changed paths: M /trunk/raw1394.c Merge change from Vger CVS, no need to lock_kernel around raw1394_release ------------------------------------------------------------------------ rev 369: bencollins | 2002-01-17 09:26:47 +0100 (Thu, 17 Jan 2002) | 2 lines Changed paths: M /trunk/ohci1394.c M /trunk/pcilynx.c Merge change from Vger CVS for __devexit_p() ------------------------------------------------------------------------ rev 368: bencollins | 2002-01-15 22:38:06 +0100 (Tue, 15 Jan 2002) | 2 lines Changed paths: M /trunk/nodemgr.c Add procfs support, based on patch from Mace Gael ------------------------------------------------------------------------ rev 367: jamesg | 2002-01-15 10:07:03 +0100 (Tue, 15 Jan 2002) | 4 lines Changed paths: M /trunk/sbp2.c Added one final (hopefully) scsi inquiry data fix from HJ Lu. One place where CMD_DMA_SINGLE and CMD_DMA_PAGE were mixed up. Also added in compatibility for 2.4.12 and earlier. ------------------------------------------------------------------------ rev 366: bencollins | 2002-01-14 22:26:42 +0100 (Mon, 14 Jan 2002) | 2 lines Changed paths: M /trunk/sbp2.c Add a little more info to the proc info output ------------------------------------------------------------------------ rev 365: bencollins | 2002-01-14 21:53:08 +0100 (Mon, 14 Jan 2002) | 2 lines Changed paths: M /trunk/sbp2.c Make printk for node fmt match nodemgr's style ------------------------------------------------------------------------ rev 364: bencollins | 2002-01-14 21:41:30 +0100 (Mon, 14 Jan 2002) | 4 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Readd pci_map_page conversion. Make sure to check this on unmap, since cmd_dma can be from pci_map_single, or pci_map_page, depending on the scsi command sent. ------------------------------------------------------------------------ rev 363: hogsberg | 2002-01-14 17:45:37 +0100 (Mon, 14 Jan 2002) | 2 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h Fix ohci1394 iso context allocation properly. ------------------------------------------------------------------------ rev 362: jamesg | 2002-01-14 12:44:51 +0100 (Mon, 14 Jan 2002) | 16 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Following changes: - Fixed sbp2 status handling to always look for scsi status returned. This fixes problems with certain sbp2 devices, such as Iomega Peerless and Jazz drives. - Removed hack from HJ to fake inquiry data, since this is not necessary now. - Fixed request buffer length passed from scsi stack to match the value in the cdb, as the mismatch upsets some sbp2 devices (again, Iomega). - Removed one instance of pci_map_page with pci_map_single, so that the driver would compile under 2.4.10. Ben, please fix this as appropriate. =) - Removed mode_sense_hack module load option, as the problem has already been taken care of for 2.4. When we branch 2.2 compatibility, will be able to clean-up the driver significantly. - Reverted one change from HJ for max sbp2 scatter gather element size, as it must be something less than 64KB (16 bit field in ORB structure). ------------------------------------------------------------------------ rev 361: dmaas | 2002-01-12 11:54:18 +0100 (Sat, 12 Jan 2002) | 4 lines Changed paths: M /trunk/dv1394.c - applied (slightly modified) patch from Ben Collins to use sg.page instead of sg.address when specifying the DMA scatterlist ------------------------------------------------------------------------ rev 360: bencollins | 2002-01-12 07:02:02 +0100 (Sat, 12 Jan 2002) | 12 lines Changed paths: M /trunk/nodemgr.c M /trunk/nodemgr.h M /trunk/sbp2.c M /trunk/sbp2.h - Commit patch from HJ to better handle nodemgr probe. Use it in sbp2. - In sbp2, use the vendor_id/module_id strings for the SCSI layer. - Make nodemgr startup a thread for each host card to handle probes. Don't schedule each node through keventd now, just perform the checks in the hosts thread. This atleast keeps us from bottle-necking keventd, and adds parallelism on a per host basis (TODO: parallelize each node probe). - SBP2, add compatibility for changes in 2.5.x. Most noticably, "use_new_eh_code" is assumed in 2.5.x. Also, use scsi_{un,}register_host for 2.5.0+ instead of scsi_{un,}register_module. And finally, use pci_map_page() for scatterlist (IOW, merge DaveM's 2.5.x tree changes). ------------------------------------------------------------------------ rev 359: bencollins | 2002-01-12 06:56:38 +0100 (Sat, 12 Jan 2002) | 2 lines Changed paths: M /trunk/highlevel.c M /trunk/ieee1394_transactions.c M /trunk/pcilynx.c M /trunk/video1394.c Use %s for __FUNCTION__ to avoid gcc-3.0.3 warnings ------------------------------------------------------------------------ rev 358: bencollins | 2002-01-12 06:54:52 +0100 (Sat, 12 Jan 2002) | 2 lines Changed paths: M /trunk/ohci1394.c Use %s for __FUNCTION__ to avoid gcc-3.0.3 warnings. Also add new cf_put_str function for config rom generation, so we don't have to use the cf_put_4bytes crap. ------------------------------------------------------------------------ rev 357: bencollins | 2002-01-12 06:53:33 +0100 (Sat, 12 Jan 2002) | 2 lines Changed paths: M /trunk/ieee1394_core.h Use the new minor() macro for 2.5.x compatibility. ------------------------------------------------------------------------ rev 356: bencollins | 2002-01-12 06:52:58 +0100 (Sat, 12 Jan 2002) | 2 lines Changed paths: M /trunk/ieee1394_core.c Use %s for __FUNCTION__ to avoid gcc-3.0.3 warnings. Also, use the new minor() macro for 2.5.x compatibility. ------------------------------------------------------------------------ rev 355: bencollins | 2002-01-12 06:51:40 +0100 (Sat, 12 Jan 2002) | 2 lines Changed paths: M /trunk/ieee1394_types.h Add compatibility macro for s/MINOR/minor/ on 2.5.x ------------------------------------------------------------------------ rev 354: jamesg | 2002-01-08 13:34:06 +0100 (Tue, 08 Jan 2002) | 4 lines Changed paths: M /trunk/sbp2.c Remove io_request_lock when completing scsi commands now that we're running as an enhanced scsi driver (no longer uses old scsi_done routine). Have tested this against an SMP 2.4 kernel, but not a true SMP machine. ------------------------------------------------------------------------ rev 353: jamesg | 2002-01-07 13:04:10 +0100 (Mon, 07 Jan 2002) | 9 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Minor fixes/enhancements: - Enable use_new_eh_code for scsi stuff - Do not write all ones for NULL ORB high/low fields, but rather leave reserved areas zeroed (per SBP2 spec) - Use newer scsi transfer direction passed down instead of our internal direction table - Bumped login time-out to 20 seconds, as some devices are slow - Fixed a couple scsi unregister bugs on module unload ------------------------------------------------------------------------ rev 352: hogsberg | 2002-01-07 02:11:29 +0100 (Mon, 07 Jan 2002) | 2 lines Changed paths: M /trunk/video1394.c Added flexible iso dma context handling from Kurt Konolige. ------------------------------------------------------------------------ rev 351: hogsberg | 2002-01-07 02:00:07 +0100 (Mon, 07 Jan 2002) | 2 lines Changed paths: M /trunk/ohci1394.c Fix mismatch in context number and register when allocating iso dma contexts. ------------------------------------------------------------------------ rev 350: hogsberg | 2002-01-04 05:10:17 +0100 (Fri, 04 Jan 2002) | 3 lines Changed paths: M /trunk/ohci1394.c Remove obsolete usage of dma_cache_wback_inv. We use pci_alloc_consistent so there is no use for dma_cache_wback_inv. Spotted by Atsushi Nemoto. ------------------------------------------------------------------------ rev 349: hogsberg | 2002-01-04 04:32:25 +0100 (Fri, 04 Jan 2002) | 2 lines Changed paths: M /trunk/highlevel.c Fix address range bug reported by Steve Kinneberg. ------------------------------------------------------------------------ rev 348: bencollins | 2001-12-29 22:45:21 +0100 (Sat, 29 Dec 2001) | 2 lines Changed paths: M /trunk/sbp2.c Lock io_request_lock to prevent lockups on SMP ------------------------------------------------------------------------ rev 347: bencollins | 2001-12-29 22:32:48 +0100 (Sat, 29 Dec 2001) | 2 lines Changed paths: M /trunk/sbp2.c Add a kbuild-2.5 wrap around some scsi includes. ------------------------------------------------------------------------ rev 346: ddennedy | 2001-12-11 11:55:18 +0100 (Tue, 11 Dec 2001) | 2 lines Changed paths: M /trunk/Config.in M /trunk/dv1394-private.h M /trunk/dv1394.c M /trunk/dv1394.h M /trunk/ieee1394_core.c M /trunk/ieee1394_core.h proc_fs for ieee1394, dv1394 ------------------------------------------------------------------------ rev 345: hogsberg | 2001-12-11 04:18:07 +0100 (Tue, 11 Dec 2001) | 3 lines Changed paths: M /trunk/hosts.c M /trunk/hosts.h M /trunk/ieee1394_core.c M /trunk/ohci1394.c M /trunk/pcilynx.c Split hpsb_add_host into hpsb_alloc_host (alloc and initialize) and hpsb_add_host (just adding a host). ------------------------------------------------------------------------ rev 344: ddennedy | 2001-12-04 12:11:21 +0100 (Tue, 04 Dec 2001) | 2 lines Changed paths: M /trunk/dv1394.c POLLIN fix ------------------------------------------------------------------------ rev 343: ddennedy | 2001-12-04 11:35:20 +0100 (Tue, 04 Dec 2001) | 2 lines Changed paths: M /trunk/dv1394.c POLLIN ------------------------------------------------------------------------ rev 342: ddennedy | 2001-12-04 10:48:11 +0100 (Tue, 04 Dec 2001) | 2 lines Changed paths: M /trunk/dv1394.c remove C++ style comments ------------------------------------------------------------------------ rev 341: bencollins | 2001-12-04 02:33:39 +0100 (Tue, 04 Dec 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Handle self_id_errors count properly. Memset ohci struct after kmalloc. Handle RESET_BUS devctl properly. ------------------------------------------------------------------------ rev 340: dmaas | 2001-11-29 11:12:06 +0100 (Thu, 29 Nov 2001) | 4 lines Changed paths: M /trunk/dv1394.c M /trunk/dv1394.h M /trunk/ieee1394_core.c - fix ieee1394_dispatch_open when ieee1394.o is not a module - remove ZAP_MODULE from dv1394; it doesn't make sense for the non-modular case ------------------------------------------------------------------------ rev 339: bencollins | 2001-11-29 00:17:57 +0100 (Thu, 29 Nov 2001) | 2 lines Changed paths: A /trunk/Makefile.in Add makefile for KBuild 2.5 ------------------------------------------------------------------------ rev 338: bencollins | 2001-11-28 23:17:20 +0100 (Wed, 28 Nov 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c Use V22_COMPAT_MOD_DEC_USE_COUNT for decrementing THIS_MODULE instead of doing it explicitly ------------------------------------------------------------------------ rev 337: bencollins | 2001-11-28 23:12:44 +0100 (Wed, 28 Nov 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c We have to allow for NULL module pointers because not everything will be compiled as a module ------------------------------------------------------------------------ rev 336: bencollins | 2001-11-28 21:10:26 +0100 (Wed, 28 Nov 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Fix scope of "struct node_entry *ne" in nodemgr_remove_host() to remove compile warnings ------------------------------------------------------------------------ rev 335: dmaas | 2001-11-28 16:19:07 +0100 (Wed, 28 Nov 2001) | 6 lines Changed paths: M /trunk/Config.in M /trunk/Makefile A /trunk/dv1394-private.h A /trunk/dv1394.c A /trunk/dv1394.h Merge of dv1394 branch, part 4 of 4: - added dv1394.c, dv1394.h, and dv1394-private.h - added dv1394 into Config.in and Makefile ------------------------------------------------------------------------ rev 334: dmaas | 2001-11-28 16:04:27 +0100 (Wed, 28 Nov 2001) | 17 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/ieee1394_core.h M /trunk/raw1394.c M /trunk/video1394.c Merge of dv1394 branch, part 3 of 4: - character device dispatching: - added dispatching interface to ieee1394_core - adapted raw1394 and video1394 to the new interface NOTE! I switched video1394 to its new character device block (171,16 through 171,31). This change is necessary because major 172 has been allocated to the Moxa Intellio serial card driver. Users of video1394 will need to remake /dev/video1394 as follows: rm /dev/video1394 mknod /dev/video1394 c 171 16 raw1394 remains on major 171, minor 0. ------------------------------------------------------------------------ rev 333: dmaas | 2001-11-28 15:18:20 +0100 (Wed, 28 Nov 2001) | 6 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h M /trunk/video1394.c Merge of dv1394 branch, part 2 of 4: - add usage bits for ohci1394's IR and IT DMA contexts so that ohci1394, video1394, and dv1394 don't all try to use the same context at once. ------------------------------------------------------------------------ rev 332: dmaas | 2001-11-28 14:42:01 +0100 (Wed, 28 Nov 2001) | 7 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h M /trunk/video1394.c Merge of dv1394 branch, part 1 of 4: - add opaque user data pointer to ohci1394's IRQ hook - allow several ohci1394 IRQ hooks, so that video1394 and dv1394 can be used together ------------------------------------------------------------------------ rev 331: aeb | 2001-11-28 06:01:02 +0100 (Wed, 28 Nov 2001) | 3 lines Changed paths: M /trunk/ieee1394.h Put parentheses around negative number constants in #defines. (From big kernel cleanup patch by Nathan Myers.) ------------------------------------------------------------------------ rev 330: aeb | 2001-11-26 09:10:06 +0100 (Mon, 26 Nov 2001) | 22 lines Changed paths: M /trunk/csr.c M /trunk/highlevel.c M /trunk/highlevel.h M /trunk/hosts.c M /trunk/hosts.h M /trunk/ieee1394_core.c M /trunk/ieee1394_core.h M /trunk/nodemgr.c M /trunk/ohci1394.c M /trunk/ohci1394.h M /trunk/pcilynx.c M /trunk/pcilynx.h M /trunk/raw1394.c M /trunk/video1394.c Merge of the PCI related changes from the newPCI branch: - Make driver loading and device detection independent. hpsb_hosts can be added and removed dynamically. - Adapt hardware drivers to new PCI hotplug scheme introduced in Linux 2.4 and make use of hosts.c changes. - Divide hpsb_template into hpsb_host_driver and hpsb_host_operations structures. - hpsb_{inc,dec}_host_usage() renamed to hpsb_(un)ref_host(). They can now no longer guarantee that hosts stay loaded (think user operated hotplug hardware like PC Card) but simply try their best (increase module reference count). The only guarantee now is that the hpsb_host pointer stays valid. - Use pci_{set,get}_drvdata() in the PCILynx driver instead of managing a list of our own for cards. Version right before this merge available by tag before-newPCI-merge. ------------------------------------------------------------------------ rev 329: hogsberg | 2001-11-23 19:12:39 +0100 (Fri, 23 Nov 2001) | 2 lines Changed paths: M /trunk/sbp2.c Add 2.2 proc entries (again). ------------------------------------------------------------------------ rev 328: hogsberg | 2001-11-20 05:54:49 +0100 (Tue, 20 Nov 2001) | 3 lines Changed paths: M /trunk/linux22compat.h M /trunk/sbp2.c - Compile fix for linux 2.2 - Remove sbp2_host_count, it's available as scsi_driver_template.present ------------------------------------------------------------------------ rev 327: hogsberg | 2001-11-17 22:21:11 +0100 (Sat, 17 Nov 2001) | 2 lines Changed paths: M /trunk/hosts.c Run callbacks before shutting down card. ------------------------------------------------------------------------ rev 326: hogsberg | 2001-11-17 22:16:10 +0100 (Sat, 17 Nov 2001) | 18 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Various bugfixes/cleanups: - Remember to logout of device in sbp2_disconnect. - If we fail to reconnect to a device after bus reset remember to release unit directory, so the ieee1394 knows we no longer manage it. - Unregister scsi hosts in sbp2_remove_host when a hpsb_host goes away. - Remove stupid hack in sbp2_remove_host. - Switched to "manual" module initialization (i.e. not scsi_module.c) and moved sbp2_cleanup moved sbp2scsi_release to sbp2_module_ext. The release function is called once pr. registered scsi host, but sbp2_cleanup should only be called upon module unload. Moved much initialization from sbp2scsi_detect to sbp2_module_init. ------------------------------------------------------------------------ rev 325: hogsberg | 2001-11-17 04:38:51 +0100 (Sat, 17 Nov 2001) | 2 lines Changed paths: M /trunk/video1394.c Use list_for_each_safe when deleting from the list. ------------------------------------------------------------------------ rev 324: bencollins | 2001-11-11 20:16:37 +0100 (Sun, 11 Nov 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Better Pismo handling patch from Benjamin Herren. ------------------------------------------------------------------------ rev 323: bencollins | 2001-11-08 03:09:58 +0100 (Thu, 08 Nov 2001) | 3 lines Changed paths: M /trunk/nodemgr.c M /trunk/sbp2.c nodemgr.c: Small cleanup and list_for_each fix from Rik van Riel sbp2.c: list_for_each fix from Rik van Riel ------------------------------------------------------------------------ rev 322: hogsberg | 2001-11-08 02:22:03 +0100 (Thu, 08 Nov 2001) | 4 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/nodemgr.c M /trunk/ohci1394.c Revert delay after LPS to simple (but working) mdelay. Remove ipv4 and ipv6 entries from ohci config rom. Skip over local node in nodemgr probe code. ------------------------------------------------------------------------ rev 321: bencollins | 2001-10-21 11:06:46 +0200 (Sun, 21 Oct 2001) | 2 lines Changed paths: M /trunk/video1394.c M /trunk/video1394.h poll functionality from Kurt Konolige ------------------------------------------------------------------------ rev 320: bencollins | 2001-10-21 08:58:03 +0200 (Sun, 21 Oct 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Return after rescheduling in nodemgr_node_probe_cleanup ------------------------------------------------------------------------ rev 319: bencollins | 2001-10-19 21:50:28 +0200 (Fri, 19 Oct 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Convert nodemgr probes to be in parallel for all nodes ------------------------------------------------------------------------ rev 318: bencollins | 2001-10-19 21:49:40 +0200 (Fri, 19 Oct 2001) | 2 lines Changed paths: M /trunk/sbp2.c Fix max payload format for printk ------------------------------------------------------------------------ rev 317: bencollins | 2001-10-19 13:12:17 +0200 (Fri, 19 Oct 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Use $Revision$ in version string ------------------------------------------------------------------------ rev 316: bencollins | 2001-10-16 12:54:39 +0200 (Tue, 16 Oct 2001) | 2 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Convert to new speed/packet detection code. Also, make the use of the SBP2_{INFO,DEBUG,ERR} more consistent ------------------------------------------------------------------------ rev 315: bencollins | 2001-10-16 12:41:35 +0200 (Tue, 16 Oct 2001) | 2 lines Changed paths: M /trunk/ieee1394_types.h M /trunk/linux22compat.h M /trunk/nodemgr.c Move NODE_FMT macros from nodemgr.c to ieee1394_types.h. Move some 2.4.x compat macros from linux22compat to ieee1394_types.h. ------------------------------------------------------------------------ rev 314: bencollins | 2001-10-16 12:38:30 +0200 (Tue, 16 Oct 2001) | 2 lines Changed paths: M /trunk/ieee1394.h M /trunk/ieee1394_core.c Add new function hpsb_speedto_str(), which converts a speed code to a string suitable for printk ------------------------------------------------------------------------ rev 313: bencollins | 2001-10-14 09:09:24 +0200 (Sun, 14 Oct 2001) | 2 lines Changed paths: M /trunk/linux22compat.h MODULE_LICENSE came around 2.4.12 ------------------------------------------------------------------------ rev 312: bencollins | 2001-10-14 09:07:38 +0200 (Sun, 14 Oct 2001) | 2 lines Changed paths: M /trunk/sbp2.c Add MODULE_LICENSE ------------------------------------------------------------------------ rev 311: hogsberg | 2001-10-13 01:43:19 +0200 (Sat, 13 Oct 2001) | 2 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Fix transfer direction for scsi opcode 0x54 (send opc). ------------------------------------------------------------------------ rev 310: bencollins | 2001-10-04 23:44:15 +0200 (Thu, 04 Oct 2001) | 4 lines Changed paths: M /trunk/Makefile M /trunk/ieee1394_core.c D /trunk/ieee1394_syms.c Add ohci1394 as an exported symbol object. Move ieee1394 symbols to ieee1394_core.c and remove ieee1394_syms.c. This way things link properly on linux-2.2.x. ------------------------------------------------------------------------ rev 309: bencollins | 2001-10-02 10:23:00 +0200 (Tue, 02 Oct 2001) | 2 lines Changed paths: M /trunk/ieee1394_syms.c M /trunk/linux22compat.h M /trunk/ohci1394.c M /trunk/pcilynx.c M /trunk/raw1394.c M /trunk/video1394.c Merge in MODULE_LICENSE changes from Linus' tree ------------------------------------------------------------------------ rev 308: hogsberg | 2001-09-26 00:19:49 +0200 (Wed, 26 Sep 2001) | 2 lines Changed paths: M /trunk/pcilynx.c Change the last few packet type enum renames. ------------------------------------------------------------------------ rev 307: hogsberg | 2001-09-25 23:12:46 +0200 (Tue, 25 Sep 2001) | 2 lines Changed paths: M /trunk/hosts.c M /trunk/hosts.h M /trunk/ieee1394_types.h Convert hosts.[ch] to use struct list_head. ------------------------------------------------------------------------ rev 306: bencollins | 2001-09-19 08:20:45 +0200 (Wed, 19 Sep 2001) | 2 lines Changed paths: M /trunk/highlevel.c Convert the multiplexer to list_for_each. This is somewhat of a merge from the Linus tree ------------------------------------------------------------------------ rev 305: bencollins | 2001-09-19 08:10:23 +0200 (Wed, 19 Sep 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/ieee1394_core.h M /trunk/ieee1394_transactions.c M /trunk/ohci1394.c M /trunk/pcilynx.c M /trunk/raw1394.c M /trunk/sbp2.c Fix packet state and type enums to be less ambiguos ------------------------------------------------------------------------ rev 304: bencollins | 2001-09-19 08:04:52 +0200 (Wed, 19 Sep 2001) | 4 lines Changed paths: M /trunk/Makefile Remove duplicate rules for 2.2 kernels, and use the compatibility rules from the USB Makefile instead. Means only one set of rules to maintain, and compatibility is handled automatically. ------------------------------------------------------------------------ rev 303: hogsberg | 2001-09-16 21:55:39 +0200 (Sun, 16 Sep 2001) | 6 lines Changed paths: M /trunk/ieee1394_hotplug.h M /trunk/nodemgr.c M /trunk/nodemgr.h M /trunk/sbp2.c M /trunk/sbp2.h Remove sbp2 detection code, instead subscribe to the nodemgr driver management interface. This also removes the initial bus scan stuff since the nodemgr calls sbp2_probe for each sbp2 device already on the bus, when we register our driver. This change automtically adds hotplug support to the driver. ------------------------------------------------------------------------ rev 302: bencollins | 2001-09-16 03:33:17 +0200 (Sun, 16 Sep 2001) | 2 lines Changed paths: M /trunk/ieee1394_transactions.c hpsb_make_packet(): Only memcpy the buffer if it is non-NULL ------------------------------------------------------------------------ rev 301: hogsberg | 2001-09-16 03:00:50 +0200 (Sun, 16 Sep 2001) | 5 lines Changed paths: M /trunk/sbp2.c Change all interruptible_sleep_on to just sleep_on, since the call to call_usermodehelper in nodemgr.c eventuallly delivers a SIGCHLD to keventd, which causes login to fail. We cant really deal with signals reasonably anyway. ------------------------------------------------------------------------ rev 300: ddennedy | 2001-09-15 10:23:08 +0200 (Sat, 15 Sep 2001) | 2 lines file dv1394.c was initially added on branch dv1394-branch. ------------------------------------------------------------------------ rev 299: ddennedy | 2001-09-15 10:23:08 +0200 (Sat, 15 Sep 2001) | 2 lines file dv1394.h was initially added on branch dv1394-branch. ------------------------------------------------------------------------ rev 298: ddennedy | 2001-09-15 10:23:08 +0200 (Sat, 15 Sep 2001) | 2 lines file dv1394-private.h was initially added on branch dv1394-branch. ------------------------------------------------------------------------ rev 297: bencollins | 2001-09-14 09:36:02 +0200 (Fri, 14 Sep 2001) | 2 lines Changed paths: M /trunk/nodemgr.c M /trunk/nodemgr.h Put extra unit directory key/value pairs into an array for special usage (e.g. sbp2) ------------------------------------------------------------------------ rev 296: bencollins | 2001-09-14 09:32:24 +0200 (Fri, 14 Sep 2001) | 2 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Lots of comment cleanups. Also, convert to using nodemgr's node and unit directory entries, instead of self-probing. This is step one to converting to the hotplug scheme ------------------------------------------------------------------------ rev 295: bencollins | 2001-09-13 20:11:31 +0200 (Thu, 13 Sep 2001) | 2 lines Changed paths: M /trunk/sbp2.c M /trunk/sbp2.h Remove pismo workaround. Also, remove kernel thread code for reset handler, and use schedule_task() instead. ------------------------------------------------------------------------ rev 294: bencollins | 2001-09-13 10:51:38 +0200 (Thu, 13 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Fix little endian compile ------------------------------------------------------------------------ rev 293: bencollins | 2001-09-13 07:00:11 +0200 (Thu, 13 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Cleanup busreset woraround ------------------------------------------------------------------------ rev 292: bencollins | 2001-09-13 06:53:48 +0200 (Thu, 13 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c More ifdef's for busreset workaround. This is temporary until the issue is resolved ------------------------------------------------------------------------ rev 291: bencollins | 2001-09-13 06:35:40 +0200 (Thu, 13 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h Cleanup byteswap declerations ------------------------------------------------------------------------ rev 290: bencollins | 2001-09-13 05:28:21 +0200 (Thu, 13 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Add pismo updates from BenH ------------------------------------------------------------------------ rev 289: hogsberg | 2001-09-12 22:52:39 +0200 (Wed, 12 Sep 2001) | 4 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h Mask out busReset interrupt during selfID phase, as suggested by Kent B. Larsen. Also, remove unused field NumBusResets from struct ti_ohci. ------------------------------------------------------------------------ rev 288: bencollins | 2001-09-11 08:42:57 +0200 (Tue, 11 Sep 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Reduce hotplug output when debug is not enabled. Also, make the debug format consistent ------------------------------------------------------------------------ rev 287: bencollins | 2001-09-11 07:04:36 +0200 (Tue, 11 Sep 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Use 016 with %Lx format of guid ------------------------------------------------------------------------ rev 286: bencollins | 2001-09-11 06:53:52 +0200 (Tue, 11 Sep 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Cast guid to long long unsigned for %Lx ------------------------------------------------------------------------ rev 285: bencollins | 2001-09-11 06:29:59 +0200 (Tue, 11 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c set_phy_reg(): Initialize value of 'r' to remove compiler warnings ------------------------------------------------------------------------ rev 284: bencollins | 2001-09-11 06:28:47 +0200 (Tue, 11 Sep 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Fix nodemgr_call_policy() args when HOTPLUG not enabled ------------------------------------------------------------------------ rev 283: bencollins | 2001-09-11 06:26:23 +0200 (Tue, 11 Sep 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Fix nodemgr_call_policy() call to HPSB_DEBUG() when HOTPLUG not enabled ------------------------------------------------------------------------ rev 282: hogsberg | 2001-09-09 05:43:16 +0200 (Sun, 09 Sep 2001) | 3 lines Changed paths: M /trunk/ohci1394.c Cleanup: dont call dma_trm_reset directly in irq_handler; hpsb_bus_reset calls devctl CANCEL_REQUESTS, which does the job. ------------------------------------------------------------------------ rev 281: hogsberg | 2001-09-09 00:00:42 +0200 (Sun, 09 Sep 2001) | 3 lines Changed paths: M /trunk/ohci1394.c M /trunk/ohci1394.h M /trunk/video1394.c Reenable irq generation on iso receive. This patch also introduces symbolic names for the DMA control flags. ------------------------------------------------------------------------ rev 280: bencollins | 2001-09-08 00:42:45 +0200 (Sat, 08 Sep 2001) | 2 lines Changed paths: M /trunk/pcilynx.c Fix module load/unload, from Manfred Weihs ------------------------------------------------------------------------ rev 279: bencollins | 2001-09-08 00:41:52 +0200 (Sat, 08 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Move FAIL() macros down to where it is used ------------------------------------------------------------------------ rev 278: bencollins | 2001-09-07 04:01:38 +0200 (Fri, 07 Sep 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Make sure to set in_bus_reset when we init the card. From Todd Reichert ------------------------------------------------------------------------ rev 277: hogsberg | 2001-09-07 00:49:05 +0200 (Fri, 07 Sep 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/ieee1394_hotplug.h M /trunk/ieee1394_syms.c M /trunk/nodemgr.c M /trunk/nodemgr.h Added protocol driver management and call to usermode policy helper. ------------------------------------------------------------------------ rev 276: hogsberg | 2001-09-05 23:28:57 +0200 (Wed, 05 Sep 2001) | 2 lines Changed paths: M /trunk/sbp2.c mask out the lower 24 bits of sw_version ------------------------------------------------------------------------ rev 275: bencollins | 2001-09-05 22:43:29 +0200 (Wed, 05 Sep 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c handle_incoming_packet() is local, so make it static ------------------------------------------------------------------------ rev 274: bencollins | 2001-09-05 22:40:40 +0200 (Wed, 05 Sep 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c Make sure we set the generation in create_reply_packet(), also make create_reply_packet() static, since it's local to this file ------------------------------------------------------------------------ rev 273: bencollins | 2001-08-22 23:26:18 +0200 (Wed, 22 Aug 2001) | 2 lines Changed paths: M /trunk/raw1394.c Fix generation assignment for ISO ------------------------------------------------------------------------ rev 272: aeb | 2001-08-22 20:39:45 +0200 (Wed, 22 Aug 2001) | 2 lines Changed paths: M /trunk/raw1394.c Use user supplied generation number for packet. ------------------------------------------------------------------------ rev 271: bencollins | 2001-08-21 23:12:24 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/linux22compat.h Add sg_dma_len and sg_dma_address compat defines ------------------------------------------------------------------------ rev 270: bencollins | 2001-08-21 06:46:08 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/video1394.c Fix incorrect pointer deref ------------------------------------------------------------------------ rev 269: bencollins | 2001-08-21 06:43:54 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/raw1394.c Fix host to fi->host ------------------------------------------------------------------------ rev 268: bencollins | 2001-08-21 04:30:53 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/raw1394.c Be sure to update the packet generation before calling hpsb_send_packet() ------------------------------------------------------------------------ rev 267: bencollins | 2001-08-21 00:38:57 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/nodemgr.c Fixup pointer deref error ------------------------------------------------------------------------ rev 266: bencollins | 2001-08-21 00:31:04 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/ieee1394_hotplug.h M /trunk/nodemgr.c M /trunk/nodemgr.h M /trunk/sbp2.c Hotplug updates ------------------------------------------------------------------------ rev 265: bencollins | 2001-08-21 00:29:48 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Don't catch PhyReg interrupt ------------------------------------------------------------------------ rev 264: bencollins | 2001-08-21 00:29:03 +0200 (Tue, 21 Aug 2001) | 2 lines Changed paths: M /trunk/ieee1394_core.c M /trunk/video1394.c Fix list_for_each misusage ------------------------------------------------------------------------ rev 263: aeb | 2001-08-20 23:42:36 +0200 (Mon, 20 Aug 2001) | 3 lines Changed paths: M /trunk/raw1394.c Back out two list_for_each conversions as they are not suitable in these cases and corrupt lists in fcp_request and iso_receive in raw1394.c. ------------------------------------------------------------------------ rev 262: bencollins | 2001-08-18 08:18:50 +0200 (Sat, 18 Aug 2001) | 2 lines Changed paths: M /trunk/video1394.c Do not return after remove_card(), to make sure we get to the spin unlock ------------------------------------------------------------------------ rev 261: bencollins | 2001-08-18 01:13:27 +0200 (Sat, 18 Aug 2001) | 2 lines Changed paths: M /trunk/video1394.c Remove unsed flags var from remove_card(), and also change DMA_SPEED_XXX to SPEED_XXX ------------------------------------------------------------------------ rev 260: bencollins | 2001-08-18 00:51:51 +0200 (Sat, 18 Aug 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Fix PRINT_G macro ------------------------------------------------------------------------ rev 259: bencollins | 2001-08-17 23:53:55 +0200 (Fri, 17 Aug 2001) | 2 lines Changed paths: M /trunk/video1394.c Don't call spinlock in remove_card() since it is only called under spinlock anyway ------------------------------------------------------------------------ rev 258: bencollins | 2001-08-16 12:53:33 +0200 (Thu, 16 Aug 2001) | 2 lines Changed paths: M /trunk/nodemgr.c M /trunk/ohci1394.c M /trunk/ohci1394.h M /trunk/sbp2.c M /trunk/video1394.c Use 'long unsigned', not 'int' for spinlock flags ------------------------------------------------------------------------ rev 257: bencollins | 2001-08-16 10:47:22 +0200 (Thu, 16 Aug 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Oops, clear bus reset bit from event var, so we know it's handled ------------------------------------------------------------------------ rev 256: bencollins | 2001-08-16 10:46:24 +0200 (Thu, 16 Aug 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Cleanup interrupt handler ------------------------------------------------------------------------ rev 255: bencollins | 2001-08-14 00:41:21 +0200 (Tue, 14 Aug 2001) | 2 lines Changed paths: M /trunk/sbp2.c Merge changes from Alan Cox's tree for sleep in spinlock. Also reformat sbp2_add_host ------------------------------------------------------------------------ rev 254: bencollins | 2001-08-08 22:48:14 +0200 (Wed, 08 Aug 2001) | 2 lines Changed paths: M /trunk/nodemgr.c Only output directory info if debug is enabled ------------------------------------------------------------------------ rev 253: bencollins | 2001-08-08 21:50:09 +0200 (Wed, 08 Aug 2001) | 2 lines Changed paths: M /trunk/ohci1394.c Up version/date ------------------------------------------------------------------------ rev 252: bencollins | 2001-08-08 21:48:48 +0200 (Wed, 08 Aug 2001) | 2 lines Changed paths: M /trunk/nodemgr.c M /trunk/raw1394.c M /trunk/sbp2.c M /trunk/video1394.c Make some globals static to avoid conflicts ------------------------------------------------------------------------ rev 251: aeb | 2001-08-06 21:08:23 +0200 (Mon, 06 Aug 2001) | 2 lines Changed paths: A /trunk/ieee1394_hotplug.h Forgot to add ieee1394_hotplug.h. ------------------------------------------------------------------------